A Parallel Multimodal Optimization Algorithm for Simulation-Based Design of Power Systems

This paper proposes a parallel multimodal optimization algorithm that is combined with electromagnetic transient simulation in a platform that unifies the setup, test, and execution of optimal designs for power systems. The algorithm speeds up the design of power systems as its computations can be executed independently on a highly parallelized environment. Additional speedup is achieved by using a surrogate model to estimate the objective function in regions of suspected local optima. The estimated functions can be used in the subsequent stages of post-optimization studies, such as sensitivity analyses. Comparative studies, in terms of computation time, are conducted against sequential execution of the proposed algorithm. The optimal design of a VSC-HVDC transmission is described to demonstrate the capabilities of the proposed algorithm.